Making annual indexes of health.

A discrete-state, continuous-time Markov model of health states and state transition probabilities is outlined as the basis of a health index that would reflect the annual health distribution and expected health changes of a population. The final index is an additive one, summing expected durations of the various health states over the entire population sample. The weights attached to the expected durations are derived from the instantaneous incidence rates, or intensity functions, that define the transition probabilities. A general procedure for collecting data during overlapping 6-week periods from numerous population subsamples is described.